ABSTRACT
The Newton-Raphson (NR) algorithm for calculating d.c. responses of nonlinear circuits has been outlined in Algorithm. This algorithm is a basic realization of the NR method and does not contain auxiliary mechanisms for efficiency improvement. The NR algorithm can be convergent very quickly, when started from a point sufficiently close to the d.c. solution. In bjt and MOS circuits a rough prediction of interterminal bias voltages, required for the selection of their starting values, can be often achieved, due to the nature of their standard nonlinearities. However, even then, some circuit configurations may cause unpredictable nonconvergence. In practice, if the starting point is not properly selected then we meet numerical difficulties in the iterative process. These may fall into four categories: divergence; limit cycles; overflow and underflow run-time errors; and extremely slow convergence. The chapter presents the basic NR procedure for d.c. nonlinear analysis and technique of NR step limiting on nonlinear elements.
